An Essential Guide on Using GPU Computing Servers

Introduction

Computing servers play a crucial role in AI research, but it is struggling for many researchers to master how to use them. We attribute most problems to their lack of basic skills in using the servers and their ignorance of how the system works. This talk provides a thorough and hands-on guide on essential techniques for using GPU computing servers. It goes through skills from using SSH for logging, to using Conda and Docker for more flexible environment management. To help our audience understand better, we provide adequate but easy-to-understand explanations of technical details. Audiences can have a chance to practice the skills on a new Tong Class server and have a Q&A session, for which on-site participation is suggested.

Speaker

Yuyang Li is a member of the Technology and Network Committee of the Tsinghua Tong Class, and the technical lead of Tong Class 2020. He is responsible for managing the GPU servers and maintaining the official website for THU Tong Class. He has worked in Linux system ops for more than five years.